<p>Drilling is a key task in exploration campaigns to characterize mineral deposits at depth. Drillcores<br>are first logged in the field by a geologist and with regards to, e.g., mineral assemblages,<br>alteration patterns, and structural features. The core-logging information is then used to<br>locate and target the important ore accumulations and select representative samples that are<br>further analyzed by laboratory measurements (e.g., Scanning Electron Microscopy (SEM), Xray<br>diffraction (XRD), X-ray Fluorescence (XRF)). However, core-logging is a laborious task and<br>subject to the expertise of the geologist.<br>Hyperspectral imaging is a non-invasive and non-destructive technique that is increasingly<br>being used to support the geologist in the analysis of drill-core samples. Nonetheless, the<br>benefit and impact of using hyperspectral data depend on the applied methods. With this in<br>mind, machine learning techniques, which have been applied in different research fields,<br>provide useful tools for an advance and more automatic analysis of the data. Lately, machine<br>learning frameworks are also being implemented for mapping minerals in drill-core<br>hyperspectral data.<br>In this context, this work follows an approach to map minerals on drill-core hyperspectral data<br>using supervised machine learning techniques, in which SEM data, integrated with the mineral<br>liberation analysis (MLA) software, are used in training a classifier. More specifically, the highresolution<br>mineralogical data obtained by SEM-MLA analysis is resampled and co-registered<br>to the hyperspectral data to generate a training set. Due to the large difference in spatial<br>resolution between the SEM-MLA and hyperspectral images, a pre-labeling strategy is<br>required to link these two images at the hyperspectral data spatial resolution. In this study,<br>we use the SEM-MLA image to compute the abundances of minerals for each hyperspectral<br>pixel in the corresponding SEM-MLA region. We then use the abundances as features in a<br>clustering procedure to generate the training labels. In the final step, the generated training<br>set is fed into a supervised classification technique for the mineral mapping over a large area<br>of a drill-core. The experiments are carried out on a visible to near-infrared (VNIR) and shortwave<br>infrared (SWIR) hyperspectral data set and based on preliminary tests the mineral<br>mapping task improves significantly.</p>